# 10.5 The people: who is affected, and who was left out (/managing-ai-workers/governance-and-responsible-use/who-is-affected)

---
type: Document
title: "10.5 The people: who is affected, and who was left out"
description: "How to check who an AI Worker's output affects, including the people a list leaves out, and when to tell people that AI was involved."
status: stable
order: 210.5
ksor:
  owner: team:panaversity
  audience: [ public ]
  approval:
    by: process:panaversity
    at: 2026-10-08T09:07:56Z
chapter: "10"
part: II
expert_status: required
concepts: [ "10.5" ]
last_verified: 2026-10-08
sources:
  - id: v1-governance
    title: "Governance, Risk & Responsible Use: A Crash Course (Panaversity, first edition, verified 2026-10-08)"
    resource: https://agentfactory.panaversity.org/docs/governance-risk-responsible-use-crash-course
generated:
  at: 2026-10-08T09:07:56Z
  by: esl-rewrite/1.2.0+ksor.1
trust_tier: unverified
build_id: sha256:2f1e14f2a0f11dbc37eb5856ca9b7d35b4ce4894da74c3328cf61a3c17c03861
dirty: true
ksor_version: 0.0.60
---

**In everyday life.** A teacher picks a class team by asking the five tallest students to choose the players. Everyone chosen is a good player. Nobody asks about the students who were never considered.

**What you will learn.** How to check who an AI Worker's work affects, including the people a list leaves out, and when to tell people that AI was involved. You practice it in step 4 of the lab.

**Why it matters.** Checking who a list left out finds the people it harms before they are paid late or turned away.

The questions in 10.1 to 10.4 protect the organization and its information. This question is about the people the work affects. Ask:

1. Who is affected, including people who never see the output?
2. What could go wrong for them, and could they notice it or question it?
3. What would a fair process look like?
4. Must they be told that AI was involved, or would it matter to them?

The easiest harm to miss happens when an AI Worker picks a few items from a larger set, and people review only the ones it picked. Shortlists, risk flags, priority queues and early-pay lists all work this way. At Brightline, the book's example company, Dave, the controller, asked the worker for a list of vendors to pay early. He checked the 14 vendors on the list. He never saw the ones it left out. The control is to **sample what was excluded, as well as what was included**.[^v1-governance]

Bias, an unfair preference for some over others, does not only come in through data. It also comes in through the brief, the worker's written instructions.[^v1-governance] "Strategic vendors first" carried a standard that nobody had written down, and the list followed it. If you judge by size, small suppliers are rarely "strategic," so they were left off. The fix was not a better AI model. It was a brief that named Brightline's own commitment, to pay small suppliers within 15 days: suppliers past the 15 days come first, then the largest discounts.

**Disclosure** means telling people that AI was involved. Check rules first: law, policy, contract, professional standards, client terms, and an AI vendor's usage policy (10.3). If one requires disclosure, the decision is made. If none does, ask whether knowing AI was involved would change the reader's understanding of the work. For a reformatted table, usually not. For a schedule the auditors will rely on, usually yes. If knowing would matter and no rule decides it, disclose rather than hide it, as long as the case is ordinary. Hiding it is what damages trust when people find out later. Sometimes you should escalate the question instead: pass it to someone with more authority. Escalate if any of these signals appears: many people affected, serious possible harm, or a matter your team has no authority to decide. Law and employment are two examples of such matters.[^v1-governance]

[^v1-governance]: Governance, Risk & Responsible Use: A Crash Course, Panaversity, first edition.